Proceedings of LINAC 2004, Lübeck, Germany
THP66
MEASUREMENT AND CONTROL OF MICROPHONICS IN HIGH
LOADED-Q SUPERCONDUCTING RF CAVITIES*
T.L. Grimm, W. Hartung, T. Kandil, H. Khalil,
J. Popielarski, C. Radcliffe, J. Vincent, R.C. York,
Michigan State University, East Lansing, MI 48824
INTRODUCTION
Superconducting radio frequency (SRF) linacs with
light beam loading, such as the Rare Isotope Accelerator
(RIA), S-DALINAC, CEBAF upgrade, and energy
recovery linacs, operate more efficiently with loaded-Q,
QL, values greater than 107. The resulting narrow
bandwidth puts stringent limits on acceptable levels of
vibration, also called microphonics, which detune the
SRF cavities and require additional rf power to maintain
amplitude and phase.
RIA will use six-cell 805 MHz elliptical cavities for
acceleration of uranium from 88 MeV/u to 400 MeV/u
(β=v/c=0.41 to 0.72) with a final beam power of 400 kW
[1]. To cover this velocity range, three geometric β, βg,
values of 0.47, 0.61 and 0.81 are used [2,3]. A multicharge state uranium beam with three charge states
centered on 238U89+ and a total beam current, Ibeam, of 0.37
mA are accelerated in the elliptical cavity section of RIA.
A prototype RIA 805 MHz βg=0.47 cryomodule has
been tested in realistic operating conditions [4]. For
QL~107 operation of RIA, rf power requirements are
determined, and measurement and control of
microphonics have been demonstrated.
BEAM LOADING & RF REQUIREMENTS
Superconducting cavities require very little rf power to
generate the accelerating gradient, and therefore, have
intrinsic quality factors, Qo, greater than 108. Additional
rf power is required for beam acceleration and control of
microphonics. If the beam current is high, beam loading
is large enough so that the system bandwidth is much
larger than any microphonics detuning. Under these
circumstances, the rf generator power (Pg) equals the
beam power (Pbeam). But, if the cavity resonant frequency
is shifted, then additional rf power is required to maintain
amplitude and phase. The required Pg for a given beam
loading, coupler strength (Qext≅QL), and maximum
detuning (±δf) is given by [5]
Pg
Pbeam
Qbeam
2
2
1 Qbeam 
QL   δf QL  






1+
=
+
4 QL  Qbeam   ∆ beam Qbeam  


2πfU
Po
=
= Qo
I beamVa cos φs
Pbeam
∆ beam = half beam bandwidth =
f
2Qbeam
*Work supported by Michigan State University and DOE DE-FG0200ER41144
Technology, Components, Subsystems
RF Structures
Table 1: Beam loading requirements for 400 kW, 400
MeV/u uranium in RIA elliptical cavities
Type
6-cell
6-cell
6-cell
0.47
0.61
0.81
βg
5.12
8.17
13.46
Va(MV)
1660
2640
2600*
Pbeam(W)
9.1x107 9.1x107 1.4x108
Qbeam
3320
5280
5200
Pg(W)
3.0x107 3.0x107 4.7x107
QL
25
25
16
∆allowed(Hz)
*βg=0.81 decreased from the maximum value due to
transit time factor
For RIA, the rf generator requirements are chosen to be
twice the maximum beam power. From the previous
equation it can be shown that the maximum allowable
detuning occurs when QL/Qbeam=0.33 for these conditions.
The maximum allowable detuning for which amplitude
and phase can be maintained is then δf=2.8∆beam.
Therefore, the microphonics can detune the cavity over a
full bandwidth, ∆allowed=2δf=5.6∆beam, and the amplitude
and phase can still be maintained. The control bandwidth
is slightly smaller than the QL bandwidth (f/QL) due to the
generator requirements.
The design beam for the driver linac is 400 MeV/u
uranium with a total power of 400 kW. For this case, the
beam loading values are shown in Table 1 for a
synchronous phase, φs, of -30º. The accelerating voltages,
Va, correspond to peak surface electric fields of 32.5
MV/m and accelerating gradients of 10-15 MV/m.
Because the βg=0.81 cavity does not accelerate uranium at
the peak of its transit time curve, the maximum beam
power is nearly the same as the maximum value for the
The allowable detuning for RIA,
βg=0.61 cavity.
assuming a generator power that is twice the beam power,
is also shown in Table 1. Due to transit time effects, most
of the cavities do not supply the maximum power to the
beam. Therefore, most of the cavities will be able to
handle significantly higher detuning than shown here.
The measured microphonics with and without passive and
active damping on the prototype RIA cryomodule show
that the values in Table 1 are adequate for RIA.
Operation of SRF cavities with QL=3x107 is done at the
S-DALINAC in Darmstadt [6], and a QL in the low 107
range is proposed for the CEBAF upgrade [7]. Energy
recovery linacs would benefit from QL in the 108 range.
Therefore, the design bandwidth values proposed here for
RIA are consistent with the goals of other projects and
will be able to capitalize on the advances made for other
accelerators.
763
THP66
Proceedings of LINAC 2004, Lübeck, Germany
Finally, Lorentz detuning which is a shift in the cavity
frequency due to radiation pressure of the electromagnetic
field must be compensated for during rf turn on. For RIA,
a cw linac, this can be done slowly using the tuner and
will not be a critical issue.
METHODS OF MITIGATION
Vibrations and pressure fluctuations cause the cavity
frequency to shift (detune) only if the forces distort the
cavity walls. Typical sources of vibration are rotating
machinery, fluid fluctuations, and ground motion. Motors
will generate discrete vibration frequencies at harmonics
of their revolution rate. Fluid fluctuations such as boiling,
cavitation, and turbulent flow, and ground motion will
generate a broadband vibration spectrum.
There are many techniques to reduce microphonics or
otherwise mitigate their effect on an SRF cavity. The
mechanical frequencies of concern are usually low (less
than 200 Hz). A proper cavity and cryomodule design
can significantly reduce the microphonics problem. The
cavity can be made more rigid with thicker material and
gussets to reduce distortions, but the tuner will become
more complicated and heat transfer from the rf surface to
the helium bath may be degraded. Structural resonances
should be eliminated or shifted to higher frequency. As
an example, the lowest frequency transverse mechanical
mode of the 805 MHz cavities was removed with a
support at the center of the six-cell structure. Designing
the cavity and helium vessel so that distortions in regions
of high electric field are canceled by distortions in high
magnetic field regions can reduce the frequency shift due
to pressure fluctuations in the helium bath. Passive
mechanical dampers can be integrated into the cavity as
has become standard for quarter-wave resonators [8].
The choice of operating temperature can have a large
influence on microphonics excitation. Operation at 4 K
results in helium at atmospheric pressure with fluctuations
from the plant and pool boiling heat transfer with its
concomitant broadband vibration. Operation at 2 K has
sub-atmospheric liquid helium with lower disturbances
from the cryoplant, and heat transfer via conduction is
enhanced due to the unique properties of superfluid.
Another system design issue is cw versus pulsed
operation. Pulsed operation must deal with the transient
Lorentz force detuning, which is likely to dominate all
other microphonic sources.
Sources of vibration should be eliminated when
possible or isolated from the cryomodule. Vibrations will
be transmitted through the ground, pipes and fluids.
Passive and active absorbers can reduce the amplitude
before the vibrations reach the cryomodule and cavity.
Examples from other disciplines are optical benches,
telescopes, automobile suspension systems, and sound
cancellation [9]. Vibrations that reach the cryomodule
can be actively and passively damped before that motion
affects the cavity.
Even with the damping and isolation techniques listed
above, there will likely still be some cavity vibration.
There are several methods available to compensate the
764
microphonics. A fast mechanical tuner can distort the
cavity shape and cancel the frequency shift induced by the
microphonics. Another way to apply a force on the cavity
is to vary the accelerating gradient by a small but
acceptable amount and use the Lorentz force [10]. The
cavity frequency can also be shifted using a tunable
reactive element that is coupled through the input power
coupler or a dedicated fast tuner coupler [11,12].
Finally, small changes in the synchronous phase can
accommodate the microphonics by allowing the rf drive
frequency to remain on resonance, as is done with the
VCX tuner [12], with an additional correction to the
voltage amplitude given the known phase error.
MEASUREMENTS
The prototype RIA 805 MHz βg=0.47 cryomodule has
two multi-cell cavities with fixed input power couplers
and external tuners actuated by a room temperature
piezoelectric actuator [4]. The cavities are cooled by 2 K
superfluid with cryoplant temperature regulation. The
coupler strength, Qext, is varied from 6x104 to 6x109 using
an external transmission line transformer. Figure 1 shows
a cross section of the cavity with tuner and coupler.
Real-time frequency detuning measurements were
made for modulation rates from dc to greater than 1 kHz.
Figure 2 shows the microphonics measurement technique.
A stable signal generator in open loop mode is used to
drive the rf cavity, and a pickup from the cavity is mixed
with the input to generate an error signal that is
proportional to the cavity detuning.
Since QL is
adjustable, the system bandwidth is set so the peak
detuning is within the system bandwidth and generates
linear error signals. This technique gives real-time error
signals that can be analyzed or used for
feedback/feedforward.
Other methods give similar
results, but require a longer acquisition time, more stable
signal generator, and cannot be used for feedback [13].
Figure 1: Side view of cavity with room temperature tuner
and fixed input coupler.
Technology, Components, Subsystems
RF Structures
Proceedings of LINAC 2004, Lübeck, Germany
At 2 K low power measurements were made with the
liquid helium partially filling the distribution pipe as
would be done for operation. The maximum frequency
deviation using the calibrated error signal was less than
40 Hz peak-to-peak (see figure 2). The error signal was
calibrated using a known frequency shift from the signal
generator. The measured modulation spectrum was
primarily comprised of discrete Fourier components with
modulation frequencies less than 80 Hz. The main
components were at 59.5 and 59.7 Hz. They could be
clearly distinguished from 60 Hz electronic pickup, which
was negligible in these measurements.
Additional
components near 54 Hz, which is the lowest frequency
natural mode of the cavity, were measured and likely due
to other nearby motors. Using an accelerometer, the
primary sources of vibration were identified to be the two
cryoplant screw compressors (rated for 1500 hp/3560 rpm
and 250 hp/3550 rpm). The cryoplant was designed to
cool superconducting magnets so vibration was not a
consideration, and no isolation from the floor or the
piping was implemented. The motors are about 20 m
from the cryomodule test area. For RIA the cryoplant
motors will be farther away and better isolated.
20
3180704-001
40
60
no damping
damping
0
Peak-to-peak Cavity Detune (Hz)
80
Figure 2: Microphonics measurement and control
technique.
30
40
50
60
70
80
Disturbance Frequency (Hz)
Figure 3: Fourier components of cavity detuning with
and without active damping.
Technology, Components, Subsystems
RF Structures
THP66
Since the detuning spectrum is made up primarily of
discrete sinusoidal disturbances, an adaptive feedforward
scheme was used to damp individual Fourier components
[14]. The dominant microphonics signal was determined
to consist of two closely spaced sinusoidal disturbances
whose damping would be more problematic then a single
peak, so for demonstration purposes a small motor with a
Fourier component at about 57 Hz was placed on the
cryomodule to generate a known sinusoidal disturbance.
Figure 3 shows the Fourier components before and after
active compensation and shows about a factor of seven
damping of the 57 Hz component. The measured
microphonics are near required levels for RIA without
active damping, and can be brought below them using the
adaptive feedforward cancellation.
REFERENCES
[1] C.W. Leemann, “The Rare-Isotope Accelerator (RIA)
Facility Project,” in Proceedings of the XX International
Linac Conference, Monterey, California (2000).
[2] G. Ciovati et al., “Superconducting Prototype Cavities
for the Spallation Neutron Source (SNS) Project,” in
Proc. of the 2001 PAC, Chicago (2001).
[3] W. Hartung et al.,“Status Report on Multi-Cell
Superconducting Cavity Development for MediumVelocity Beams,” in Proceedings of the 2003 Particle
Accelerator Conference, Portland, Oregon (2003).
[4] T.L. Grimm et al., “Experimental Study of an 805
MHz Cryomodule for the Rare Isotope Accelerator”,
these proceedings.
[5] H. Padamsee et al., “RF superconductivity for
Accelerators”, John Wiley & Sons, p. 391 (1998).
[6] H.-D. Gräf, “Experience with Control of Frequency,
Amplitude and Phase”, in Proc. of the 5th Workshop on
RF Superconductivity, Hamburg, Germany (1992).
[7] L. Harwood, “Upgrading CEBAF to 12 GeV,” in
Proc. of the 2003 PAC, Portland, Oregon (2003).
[8] A. Facco, “Mechanical Mode Damping in
Superconducting
Low-β
Resonators,
Particle
Accelerators, vol. 61, pp. 265-278 (1998).
[9] A. Hull, C. Radcliffe, S.C. Southward, “Global Active
Noise Control of a One Dimensional Acoustic Duct Using
a Feedback Controller”, ASME J. Dynamic Systems,
Measurement and Control, Vol. 115, pp.488-494 (1993).
[10] J.R. Delayen, “Electronic Damping of Microphonics
in Superconducting Cavities”, in Proceedings of the 2001
Particle Accelerator Conference, Chicago (2001).
[11] A.K. Mitra et al., “Coarse and fine tuners for the
CERN PS 40 MHz Buncher Cavity”, in Proc. of the 1997
Particle Accelerator Conference, Vancouver (1997).
[12] R. Pardo et al., “Long-Term Operating Experience
for the ATLAS Superconducting Resonators,” in 9th
Workshop on RF Superconductivity, Santa Fe, (1999).
[13] M.P. Kelly et al., “Microphonics Measurements in
SRF Cavities for RIA”, in Proceedings of the 2003
Particle Accelerator Conference, Portland (2003).
[14] T. Kandil et al., “Adaptive Feedforward Cancellation
of Sinusoidal Disturbances in Superconducting RF
Cavities”, these proceedings.
765